Exceptions to the Rule: When to Change a Forecast

There are exceptions to every rule, and sales forecasting is no different.

We previously outlined eight best practices when forecasting sales. As with many things in life, there are exceptions. Here are times to bend the rules a little bit.

1. Switch it up: the same forecast method may not work for all products

Many vendors have one standard formula in place for their sales forecasting. But when do you need to give it extra consideration? Should you place an emphasis on recent trends? Or do you have products that perform better at certain times of year? Some of your products may be seasonal while others are consistent year-round sellers. Vary your forecasting method to change your reference period: Is last month more relevant, or would last year be better?

2. Past promotions skew the forecast

One of the most important considerations is promotions. This is a big figure that can change sales forecasts. What promotions have you offered in the past? A spike in sales due to a recent sale can skew all your calculations. If you are doing sales forecasts manually, you could substitute the higher sales figure with the average over the past few months. This can help you predict with a little more accuracy instead of blindly guessing a period of higher sales.

3. Prepare for upcoming promotions

Another exception to think about is if you are going to have future promotions. If you are expecting to hold big promotions in May and September, you can think about sales forecasting in terms of units or as a percentage increase. When holding sales, don’t forget that you will need to order additional inventory to meet customer demand on time. Schedule these increases in demand as far out as possible so that you factor in lead time to deliver sufficient stock.

4. Product launch spikes

If you launched in the past year, what did these items perform? For some merchants, the first few days or weeks after a product launch is huge in terms of the number of sales. Then things settle down or there is even a sales decline. It may be appropriate to exclude the first three to seven days after a product launch in order to get a more normalized sales velocity and not that big spike that can occur right at the beginning. This is especially applicable to retailers of fashion and electronics, where there is always buzz about the newest, latest, greatest style coming out.

5. When not to replenish

Whether you are looking at altering a sales forecast or your replenishment scenario, you may think about discontinuing an item. What is your threshold for discontinuation—typically measured units sold, profitability, or revenue generated? If a product is not meeting your standards, it may well be worth considering discontinuing it and liquidating current stock. ABC classification can help determine which items may not be worth your investment. C class items (usually the bottom 5% of total revenue) may not warrant the trouble of reordering from your vendor or storing in your warehouse. If that is the case, remove it from your replenishment scenario and take keep that in mind with sales forecasting.

6. Seasonal lead times

Sales and promotions aside, the last factor to consider is seasonal lead times. Do you have different lead times during different times of the year? Most factories in China cease production during the Chinese New Year for an entire month. For example, if you place an order in February or March, chances are very good the lead time will be weeks instead of days, or even months instead of weeks. Are there any other considerations that are seasonal that may affect lead time or days of stock? It is worth the time (and revenue) to create more than one scenario based on the lead time or days of stock settings.


These are all exceptions to keep in mind. When you do, you will be able to forecast your sales more accurately, have less overstock, and maximize your revenue. Sales and seasonality have the largest roles to play, and when added to the overall sales forecast, you will have the best information possible to predict sales and replenishment.